The UK government is placing high hopes on the potential for AI to improve productivity in public services. But how big is this potential? As Nghi Nguyen explains, new data on how public sector employees use their time and which tasks they consider unimportant suggests significant scope for improving both their productivity and their job satisfaction.
The pilot public service Time Use Study (TUS) by the Office for National Statistics provides insights into how public sector workers spend their time, as well as their views about work activities and their own productivity at work. Using an online diary tool, they recorded both work and non-work activities over two 24-hour periods on selected working days in February 2024. There were responses from 1,915 public sector workers living in Great Britain. Figure 1 shows the proportion of working hours spent on non-sector-specific activities, by all public sector workers, frontline or non-frontline workers.
Figure 1: Proportion of working hours spent on non-sector-specific activities, by all public sector workers, frontline or non-frontline workers
Source: Office for National Statistics, 2024
As part of the study, when carrying out non-sector specific activities, participants were asked to evaluate the importance of these tasks, and to describe their perceptions of the duration that each non-sector activity took as “needed more time”, “took the right amount of time” or “took too much time”. Non-sector specific activities refer to tasks that are not distinct to a certain type of public service. They include both specialised tasks that might require specialist knowledge such as data analysis and project management, as well as more general tasks such as attending meetings and events.
The results suggest a fair proportion of non-sector specific activities were seen to take “too much time” (19%), compared with 72% taking “the right amount of time”, and 8% perceived as “needed more time” (Figure 2).
Figure 2: Proportion of time spent on non-sector-specific activities, according to how long participants felt they took to complete, Great Britain 2024
Source: Office for National Statistics, 2024
There were similar results for the respondents’ views about the importance of activities; for example over 20% of time spent on meetings was considered “not important” (Figure 3).
Figure 3: Percentage of time spent on non-sector-specific activities and participants’ understanding of their importance, Great Britain 2024
Source: Office for National Statistics, 2024
Interesting in themselves, the time use results provide a fair guide to the potential for using AI to improve public sector productivity which is something the UK Government is keen to advocate. For example, within the “meetings and events” category, 23% of the total time spent on organising meetings and meeting administration was considered excessive, yet only 27% of the total time spent was rated as very important.
Participants also pointed out that attending meetings and events took too much time and was not as important as other tasks. Aside from the obvious solution of reducing unnecessary meetings, AI-enabled summarisation and transcription tools can capture key discussion points. These might, allow workers to stay informed without the need to attend every meeting – freeing up time for more critical tasks. It would certainly save on the time spent on minute-writing, or planning meetings across diaries.
Overall, AI tools can, among other things:
- draft meeting agendas based on past data and user prompts
- create meeting invitations and managing replies
- estimate event costs and conduct preliminary risk assessments
- recommend suitable setups and venues using information on attendees
- summarise, transcribe and translate meeting discussions for broader accessibility
Staff management activities, including supervision tasks, leave allocation or task delegation, have the second-highest proportion of total time spent on them that workers felt “took too much time” at 24.1%. In this category, AI tools can streamline workforce management through staff scheduling, absence management and even delegating tasks based on worker availability and skillsets. This could be particularly useful in sectors facing workforce shortages, such as the NHS.
Additionally, AI-powered tools can track task progress and send deadline reminders, replacing the need for constant supervision. AI-driven chatbots can also handle frequently asked queries regarding work procedures, contracts, and HR policies.
Accordingly, AI tools can:
- optimise staff scheduling based on availability and skillsets
- predict workforce needs using historical data, e.g. expected admission rates to hospitals
- track task progress and send automated deadline reminders
- identify areas for improvement in task efficiency
- handle employee FAQs on work procedures and policies
For most of the public services included in the study, record management accounted for a substantial share of their working hours, with time spent on these tasks ranging from 2.5% for secondary education workers to over 15% for those working in prison, probation and courts and tribunals services.
Within the record management category, public service workers reported that 41% of their time spent completing and processing forms was considered as “took too much time”, while over 18% of the time spent writing and reviewing reports was reported to “need more time”.
New AI tools based on large language models, such as ChatGPT or Claude, have the potential to process, summarise and extract information with greater speed than humans, who are prone to cognitive biases and manual errors – although the AI tools themselves are still prone to ‘hallucinations’, or errors.
In terms of record management tasks, AI tools can:
- summarise and extract information from large volumes of paperwork, reports and case files
- assist editing and reviewing documents, reports and databases by flagging inconsistencies or missing information
- automate form processing by set criteria
- manage and update bookings and appointments, e.g. allocate patient appointments based on patient and staff availability, and other data (such as when patients are most likely to miss their appointments) to minimise time waste
- cross-reference patient profiles and history to support the writing of medical notes and treatment plans
The potential for AI to contribute to improvements in public service productivity is clear – and could also improve the quality of the jobs. For example, England has faced a persistent challenge in teacher recruitment and retention. In the 2022 Working Lives of Teachers survey, 80% of teachers and school leaders reported high workload as a reason for leaving the state education sector. Many teachers also indicated that general administration tasks, lesson planning and data recording, input and analysis activities consumed too much of their time.
By integrating AI to handle routine paperwork and repetitive administrative tasks, public sector workers—especially frontline staff—can focus more on their core responsibilities, improving their job satisfaction, work-life balance, and efficiency.
But will this degrade the quality of work? A study in 2023 found that even linguists and reviewers from top journals were mostly unsuccessful at distinguishing AI- and human-generated writing. The quality of AI output is still questionable but it is undeniable that AI tools are capable of performing the more methodical and repetitive administrative tasks that are taking up a large chunk of worktime of public service workers. The new public sector time-use data helps identify which tasks can be automated, and the scope of the work suitable for the use of AI tools that are already available.
With that being said, AI will never be a replacement for human workers, particularly in the public sector, where case-by-case judgment and human oversight are essential in complex and high-impact services. Instead, AI can alleviate administrative burdens, allowing public service workers to focus on meaningful tasks that require critical thinking, problem-solving, and direct engagement with the public. By streamlining processes which this new data suggests take up about one-fifth of public sector employees’ time, AI has the potential to enhance efficiency, reduce burnout, and improve work-life balance for those in these essential roles.
The views and opinions expressed in this post are those of the author(s) and not necessarily those of the Bennett Institute for Public Policy.
